A ubiquitous element of current fixed service satellite repeaters is the output multiplexer (also called “mux”). An output multiplexer filters the individual signals received from multiple high power amplifiers and combines them into a composite waveform that is routed to the antenna beam formers via a single transmission line. FIG. 1 illustrates a conventional output multiplexer 5 and shows the filters 7, comprised of resonant structures, and the manifold 9 into which signals are injected and combined. Of special note is that the filters 7 interface directly with the manifold 9, without any intermediate provision to isolate the filter function from the combining function. This form achieves considerable economies of size and power efficiency, but results in a highly complex design that must be optimized and aligned as a whole because of the extreme interdependence of all constituent parts. Accordingly, output multiplexers are inherently sensitive structures.
Dimensional stability is paramount to the proper functioning of an output multiplexer. A dimensional change in the resonant structure of a filter, due to thermal expansion, alters the passband frequency. Changes in manifold dimensions degrade the filter performance because of the skewed match. Output multiplexers have been traditionally fabricated from very low expansion steel alloys of which Invar, with a coefficient of thermal expansion (CTE) near 1 part per million per Celsius degree (ppm/C.°), is most common. As conventionally known, the coefficient of thermal expansion (CTE) is generally defined as the fractional increase in length per unit rise in temperature.
Two substantial commercial forces are influencing the design of output multiplexers. First, increasing traffic volume is necessitating maximum use of the available radio spectrum. A high power signal incident on the band edge of a filter represents a potentially damaging fault condition, therefore, any uncertainty in the location of the edges due to filter drift renders that part of the passband unusable. Second, high traffic densities and/or direct broadcast applications require increased power levels within output multiplexers, creating ever harsher thermal environments.
In the face of these trends, even the modest expansion of Invar equipment begs improvement. However, with currently employed power levels upwards of 450 Watts per channel, the design space becomes severely constrained. Invar exhibits poor thermal conduction properties, which lead to self-defeating high temperatures. Temperatures of some extant designs approach the limits of the output multiplexer materials. Alternate low CTE materials, such as carbon fiber composites, share this conduction deficiency. Additionally, Invar has undesirably high mass density. Aluminum is a preferred material in general spacecraft application because of its lightness, strength, and excellent thermal conductivity. However, aluminum also has a noteably high CTE of 23.4 ppm/C.°, which is untenable in a conventional output multiplexer application.
Contending with the heightened thermal flux requires a superior path to a heat sink. Structural elements that support output multiplexers and sink the heat are invariably made of aluminum. Securely fixing a low coefficient of thermal expansion (CTE) output multiplexer to an aluminum support, results in intolerable stress in the presence of temperature changes. Historically, Invar output multiplexers have been mounted by means of flexible brackets that alleviate the thermal stress, but in the high power regime such necessarily minimal sections present an unacceptable heat flow bottleneck.
In view of the above-noted design constraints, an aluminum output multiplexer is highly desirable in a high power regime and is well suited in every aspect except in the dimensional stability of the radio frequency boundaries. What is needed is a means of compensating for the radio frequency effects of thermal expansion associated with an aluminum output multiplexer.
This filter compensation problem has been widely examined over the years. High power filters typically consist of free space cylindrical cavities with tuning screws that penetrate the cylinder walls for fine frequency adjustment. Proposed or embodied compensation solutions generally fall into three categories each having their own limitations.
One compensation approach is disclosed in U.S. Pat. No. 4,677,403 to Kich et al. that describes the use of multiple filter structures where the tuning screw, or similar field perturbing element, penetration or diameter varies with temperature. The wave mechanics of the resonator require that the penetration of the tuning screw reduce as the cavity temperature rises, therefore, merely selecting a material with a complimentary coefficient of thermal expansion (CTE) is not an option. These multiple filter structures typically use bimetal springs or shape memory alloys to manipulate the screw penetration. However, in very high power regimes the tuning screw itself is a locale of significant radio frequency energy dissipation and because it is small is therefore subject to large temperature change. Such local temperature may not adequately track the temperature change of the entire cavity, which is what determines the frequency behavior. Also, in dual mode cavities, individual compensating screws are required for the orthogonal modes. These features must track each other very precisely in order to preserve filter alignment, a very difficult attribute to maintain in practice.
Other compensation approaches involve deforming the end wall of a cylindrical cavity in order to change its apparent length as disclosed in U.S. U.S. Pat. No. 6,433,656 to Wolk et al., U.S. Pat. No. 6,535,087 to Fitzpatrick et al. and U.S. Pat. No. 6,002,310 to Kich et al. These variations include bimetal diaphragms or constraining devices (rings or braces) made of a contrasting CTE material that impose forces on a flexible end wall. However, these devices operate locally and respond to thermal effects in the immediate vicinity of the compensating end wall. Temperature gradients along the cavity length, which are increasingly significant at elevated power levels, are not integrated. Also, all the mechanisms realize the motive force through flexures. The features or parts that cause the compensating motion do so under bending from thermal stress. Consequently, the nature and degree of movement is highly sensitive to variabilities in the material modulus and/or the part dimensions. Interim thermal testing and adjustment are generally required. Further, flexure based mechanisms tend to create non-linear movement with respect to temperature, where a linear response is more desirable. Finally, all the present mechanisms have limitations of the range of motion available. Higher temperatures or longer cavities require increasingly long strokes of the diaphragm.
Another compensation approach addresses the distinct, but related problem of maintaining constant separation of reactive elements in a transmission line and is disclosed in U.S. Pat. No. 5,428,323 to Geissler et al. and U.S. Pat. No. 6,897,746 to Thomson et al. This compensation mechanism is based on the dispersion property of rectangular waveguide. The effective wavelength of a signal, within a rectangular waveguide, depends upon the larger “a” dimension of the waveguide such that a narrowing of the waveguide increases the wavelength of signals present. However, expansion of the manifold along its length alters the spacing between filters, which disturbs the very critical spatial separation of the channel filters. These important spatial relationships are determined by the signal phase differentials between the junctions. Increasing the wavelengths of the signals at similar rate as the manifold lengthens by thermal expansion negates the consequences of thermal expansion. This compensation is achieved by causing the narrow wall of the waveguide to bend inwards (in response to heating) or outward (in response to cooling). However, there are several limitations of this approach associated with the design challenges of a practical embodiment. The wall that must be bent is the small wall and accordingly is inherently resistant to deformation. It is difficult to compensate without excessive forces or unreasonably thin wall thickness. Also, to operate successfully, bending of the wall needs to be highly uniform over the affected length of the manifold adding to these difficulties.